Adaptive Neural Network Optimal Fault-Tolerant Trajectory Tracking Control for Underactuated Autonomous Underwater Vehicles

For the control of underactuated autonomous underwater vehicles, it is necessary to consider not only fault-tolerance but also optimal performance. Thus, an adaptive neural network-based optimal fault-tolerant trajectory tracking control method is proposed. First, the system output is redefined using a coordinate transformation method to address the underactuation issue. Subsequently, a control law integrating an adaptive disturbance observer and an improved backstepping approach is designed, achieving online estimation and compensation of composite disturbances caused by both actuator faults and external disturbances. On this basis, a radial basis function neural network is utilized to construct an optimal control law, which effectively suppresses the impact of actuator faults on the closed-loop system stability while accomplishing trajectory tracking control. Theoretical analysis demonstrates that all signals of the closed-loop system are uniformly ultimately bounded, and numerical simulations verify the superiority and feasibility of the proposed method.